Sign in Subscribe
Energy management

Energy Storage Systems Vulnerable to Cyber-Attacks

Strategic Insights

— 1 min read

Energy storage systems, particularly those with state-of-charge estimation, are highly susceptible to malicious cyber-attacks, threatening their secure operation. Researchers from North China Electric Power University have proposed a model-free scaling attack method using reinforcement learning to overcome conventional Bad Data Detection mechanisms. This approach allows attackers to bypass traditional False Data Injection Attacks, posing a significant threat to grid-connected energy storage systems.

Key Takeaways:

  • Energy storage systems are critical in power regulation and load balancing, but their state-of-charge (SOC) estimation is vulnerable to cyber-attacks.
  • The proposed model-free scaling attack method uses reinforcement learning to optimize attack strategies and maximize SOC estimation errors while remaining undetected.
  • The trained agent successfully bypasses conventional Bad Data Detection mechanisms and induces average SOC estimation errors of 9.91% and 10.60% on IEEE 14-bus and IEEE 57-bus test systems, respectively.
  • The attack method does not require full knowledge of power system topology, reducing dependency on system-specific information.
  • Simulation experiments on test systems integrated with battery energy storage systems (BESS) validate the stealth and effectiveness of the proposed approach.
  • The researchers used the Deep Deterministic Policy Gradient (DDPG) algorithm to optimize the attack strategy and learn optimal scaling factors.
  • The study is funded by the Postdoctoral Fellowship Program of the Chinese Postdoctoral Council and Fundamental Research Funds for the Central Universities.
  • Additional researchers involved in the study include Yuancheng Li and Weijia Zhao.

Statistics:

  • 9.91%: Average SOC estimation error on IEEE 14-bus test system
  • 10.60%: Average SOC estimation error on IEEE 57-bus test system
  • 134: Issue number of the Journal of Energy Storage
  • 2025: Year of publication for the study in the Journal of Energy Storage, Volume 134
  • 2: Number of researchers involved in the study, excluding the lead author

Sources:

  • A Model-free Scaling Attack Method Against State-of-charge Estimation In Battery Energy Storage Systems Based On Reinforcement Learning. Journal of Energy Storage, 2025;134.
  • NewsRx. Researchers at North China Electric Power University Release New Data on Energy Storage (A Model-free Scaling Attack Method Against State-of-charge Estimation In Battery Energy Storage Systems Based On Reinforcement Learning). Information Technology Newsweekly. November 4, 2025; p 710.
  • Journal of Energy Storage. Radarweg 29, 1043 Nx Amsterdam, Netherlands.
  • Rong Huang, North China Electric Power University, School of Control and Computer Engineering, Beinong Rd, Beijing 102206, People's Republic of China.

Read more